Patients with Central Visual Field Loss Adopt a Cautious Gait Strategy during Tasks That Present a High Risk of Falling

Abstract

Purpose.: To investigate how patients with central visual field loss (CFL) complete adaptive gait tasks when compared to visual normals and determine whether task difficulty significantly affects movement control.

Methods.: Ten patients with CFL and 12 visual normals negotiated a floor-based obstacle (of different heights, 5 and 10 cm) and also walked across an unobstructed laboratory (no obstacle present). Analysis assessed the kinematics of human movement for each task.

Results.: During obstacle crossing, compared to visual normals, patients with CFL lifted their lead and trail foot significantly higher to avoid the obstacle, reduced horizontal crossing velocity (only significant at low obstacle height), and increased head flexion to look down at more immediate areas of the ground (P < 0.05). During the walking only trials there was no significant difference between the two groups in any of the kinematic measures.

Conclusions.: Compared to visual normals, patients with CFL adopt a cautious gait strategy during tasks that present a high risk of falling, such as obstacle crossing. However, under conditions that present a low risk of tripping or falling, such as level walking, differences appear minimal.

Introduction

Falls present a significant risk of injury among the elderly population, with the risk increasing with age. ^1,2 Falls occur most commonly during adaptive gait, when negotiating steps/stairs, ^3–5 curbs, ³ and obstacles, ^1,2 and often result in head injuries ⁶ and hip fractures. ⁷ The causes of falls are multifactorial, with individuals being at a higher risk of falling if they have reduced balance or poorer vision; are depressed; or have dementia, lower systolic blood pressure, or disorders with gait/reliance upon a mobility aid(s), and take psychotropic drugs. ² Vision/visual impairment is among the most important clinical risk factor associated with falls. ⁸  

Patients with central visual field loss (CFL) have been reported to be at an increased risk of falling compared to those with normal vision, ^9–13 supporting the view that the central area of the visual field is an important predictor of mobility performance. ¹⁴ Indeed, under difficult environmental conditions (such as walking across different surface terrain or under extreme light levels) compared to visual normals, patients with CFL adopt a more cautious gait strategy through walking slower and increasing head flexion to look down at more immediate areas of the ground. ^15,16 Patients with CFL also adopt a flatter foot at contact with the ground when negotiating the different surface terrain. ¹⁵ When a population-based sample of older adults completed a mobility course (which consisted of walking through a cluttered environment with objects positioned throughout), CFL was associated with reduced walking speed and increased number of contacts with objects in the environment. ¹² However, other research found that, when completing a mobility course, there was no difference in performance (measured using similar indices as Turano et al.¹²) between patients with CFL and those with normal vision, ^13,18 or when required to walk a pre-specified distance. ¹¹ Wood et al. ¹⁹ also found that CFL had a minimal effect on gait when walking along level terrain, being associated only with the time both feet were in contact with the floor (double support time). ¹⁹ The effect that CFL has on mobility performance may be affected by task difficulty. Indeed patients with CFL have been shown to have poorer mobility performance when walking in increasingly difficult environments. ²⁰ However, since Kuyk and Elliott did not include a control group, ²⁰ it is difficult to ascertain whether the environment/terrain or visual impairment had influenced the results. 

Despite the wealth of research that has investigated the effect of CFL on patients' ability to walk a mobility course or along level terrain, ^{12,13,17–21} investigation into the effect of CFL on adaptive gait (such as obstacle crossing and step ascent) remains scarce. Indeed, only a few studies have produced a detailed kinematic assessment (through using three-dimensional motion analysis) of how CFL affects adaptive gait and compared performance to normal vision participants (Spaulding et al. assessed gait when walking across different surface terrain and under extreme light levels^15,16). Assessment of human movement using three-dimensional motion analysis is regarded as the gold standard measurement of gait ¹⁷ as it provides additional detail of a movement compared to traditional global measurements, such as “time to completion,” a measurement used commonly during obstacle course paradigms. ^{12,13,18–21} The detailed analysis offered by motion analysis systems has enabled important parameters specific to the particular movement to be analyzed, such as the height that individuals lift their foot over an obstacle during adaptive gait, a key parameter that indicates the likelihood of participants tripping and falling during obstacle crossing. It, therefore, allows a greater understanding of how CFL affects mobility/adaptive gait through providing an increasingly detailed assessment of mobility/adaptive gait among such patients. 

Currently, a precise understanding of how CFL affects adaptive gait remains scarce, and it also is unknown whether task difficulty has a significant effect on how patients with CFL complete everyday mobility tasks compared to those with normal vision. In our study, patients with CFL were tasked with completing two common everyday tasks of different difficulty; walking up to and stepping over an obstacle during ongoing gait (a task that to our knowledge has not been investigated previously), and walking across the laboratory with no obstacle present. Performance was compared to participants of a similar age with normal vision. A detailed assessment of the kinematics of obstacle crossing/level walking was achieved using a three-dimensional motion analysis system (Vicon motion analysis; Oxford Metrics Ltd, Oxford, UK). 

Methods

Participants

Ten participants with CFL (age 77 ± 10 years, height 1.64 ± 0.10 m, weight 66.6 ± 11.5 kg) and 12 participants with normal vision (age 72 ± 6 years, height 1.67 ± 0.09 m, weight 69.25 ± 10.0 kg) were recruited for the study (see Table 1 for individual participant anthropometric data). The mean differences between age, height and weight across groups were not significantly different (t-test t[20] = −0.578, P = 0.81; t[20] = −0.649, P = 0.38; and t[20] = 1.56, P = 0.21, respectively). The tenets of the Declaration of Helsinki were observed and the experiment gained approval from the University's Ethical Committee. Written informed consent was obtained from each participant before undertaking the study. 

Table 1. 

 

Anthropometric Data for CFL and Normal Vision Participants

Participant	Sex	Age	Height	Weight
CFL 1	F	80	1.55	53
CFL 2	M	67	1.64	80
CFL 3	M	82	1.7	67
CFL 4	M	57	1.84	90
CFL 5	F	71	1.73	73
CFL 6	F	84	1.56	64
CFL 7	F	81	1.59	58
CFL 8	F	82	1.54	57
CFL 9	F	92	1.7	66
CFL 10	F	78	1.54	58
Norm 1	M	67	1.84	85
Norm 2	M	75	1.67	60
Norm 3	F	76	1.63	54
Norm 4	F	82	1.54	68
Norm 5	F	71	1.65	77
Norm 6	F	72	1.62	79
Norm 7	F	82	1.53	57
Norm 8	F	73	1.68	68
Norm 9	F	71	1.65	60
Norm 10	M	67	1.75	80
Norm 11	F	67	1.75	70
Norm 12	F	61	1.67	73
CFL		77 (10)	1.64 (0.10)	70.4 (21.8)
Norm		72 (6)	1.67 (0.09)	69.25 (10.0)

 Height measured in meters (m) and weight in kilograms (kg). Norm, normal vision.

Eligibility Criteria

Health and physical fitness of all participants were assessed through a self-report questionnaire. Participants were excluded if they reported any history of neurologic or musculoskeletal disorders that could affect balance or gait; or had insulin-dependant diabetes; fallen in the last 12 months (a fall being defined as unintentionally coming to rest on the ground or some other level ²² ); vestibular disturbances; somatosensory/orthopedic problems (i.e., issues with their feet that resulted in being unstable when standing or walking); taken medication that could affect balance or vision; had cataract surgery in the last year; required a mobility aid for support; a history of eye disorders, including amblyopia, strabismus, or ocular pathology (except those that resulted in CFL in the visually impaired group); or failed to score the minimum requirement to pass the Mini Mental State Examination ²³ (MMSE). All participants were physically active and independent as determined through the activity scale of the Allied Dunbar Fitness Survey. ²⁴ Participants reported regularly undertaking light-to-moderate intensive activities, such as gardening, walking, and swimming. These inclusion criteria were based on a combination of previous research that assessed the kinematics of gait in healthy older adults. ^25,26 To determine whether there was a significant difference between the level of health and fitness between subject groups, statistical analysis was performed on the scores of participants' Allied Dunbar Fitness Survey. For each participant, a frequency count was conducted on any response to a question indicating reduced health/fitness. Results highlighted that there was no significant difference in the activity scale between visually impaired and normal vision participants (U = 33.00, z = − 1.521, P = 0.15). 

Visual Assessments

Visual examination and diagnosis of all participants was provided by a consultant ophthalmologist; all participants with CFL had diagnosed bilateral macular problems, 9 participants were diagnosed with age-related macular degeneration and 1 with a macular hole. Distance visual acuity (VA) and contrast sensitivity (CS) were measured monocularly and binocularly. However, since both eyes normally are used during adaptive gait, only binocular scores are reported. VA was measured using a Bailey-Lovie LogMAR chart at a working distance of 4 m using a letter-by-letter scoring system (0.02 LogMAR). If participants were unable to read the largest letters on the LogMAR chart at 4 m, shorter distances were used and the score adjusted accordingly. CS was measured using the Pelli-Robson chart at 1 m and scored by letter (0.05 log units). Stereopsis was measured using the Frisby stereo test initially at a distance of 60 cm. All visual measurements were performed by an optometrist under normal room illumination (531 lux) using the best-corrected spectacle prescription for that distance as determined by subjective refraction. The same level of illumination was used in the laboratory for adaptive gait. The mean binocular VA scores for CFL and normal vision groups were 0.77 ± 0.48 and −0.09 ± 0.08 logMAR, respectively. CS scores for CFL and normal vision groups were 1.09 ± 0.31 and 1.68 ± 0.06 log, respectively. From the 10 visually impaired participants recruited, only two had measurable stereoacuity (340 and >600 seconds of arc). The normal vision group recorded 55 ± 17 seconds of arc. 

Visual field assessment was conducted using a Humphrey Field Analyzer (Carl Zeiss Meditec, Inc., Dublin, CA) SITA-Standard 24-2 threshold test. All participants were tested monocularly wearing their best near correction as reported previously. ²⁷ Binocular visual fields were calculated based on the more sensitive of the two visual field locations for each eye (i.e., the “best location” model ²⁸ ). To determine the size/extent of central visual field loss among the visually impaired group, specific areas of the visual field were measured and compared to visual normals. This approach was based upon previous research that, using each patient's integrated binocular visual field plot, defined the mean threshold ^29,30 for the central 5°, central 10°, and mid-peripheral 10–30°, which identified the central 4 and 16, and outer 38 test points, respectively (Fig. 1). The mean binocular central 5°, 10°, and mid-peripheral 10–30° VF scores for CFL were 20 ± 8, 22 ± 7, and 22 ± 7 decibels (dB), respectively. The mean binocular central 5°, 10°, and mid-peripheral 10–30° VF scores for normal vision patients were 32 ± 2, 30 ± 2, and 27 ± 2 dB, respectively. Individual participant visual characteristics can be found in Table 2. 

Figure 1. 

 

Integrated binocular visual field plot for a visually normal participant with illustration of the central 5° and 10°, and mid-peripheral 10–30° grids overlaid.

Table 2. 

 

Clinical Visual Function for CFL and Normal Vision Participants, Showing CS, VA, Stereopsis, and Visual Field

Participant	VA	CS	Stereopsis	Central 5° VF	Central 10° VF	10–30° VF
CFL 1	1.14	1.05	n/a	10	12	7
CFL 2	0.48	1.15	n/a	14	21	25
CFL 3	0.44	1.2	n/a	23	24	22
CFL 4	1.16	1.25	n/a	30	30	29
CFL 5	0.00	1.7	340	33	32	30
CFL 6	1.14	0.85	n/a	14	12	15
CFL 7	0.805	0.9	n/a	18	23	25
CFL 8	1.12	1.15	n/a	23	22	18
CFL 9	0.1	1.1	>600	17	27	23
CFL 10	1.32	0.5	n/a	7	10	6
Norm 1	−0.1	1.75	40	33	32	29
Norm 2	−0.06	1.7	40	33	32	29
Norm 3	0.00	1.65	80	29	28	27
Norm 4	0.00	1.6	40	31	30	28
Norm 5	−0.16	1.75	70	31	30	29
Norm 6	−0.18	1.75	55	30	28	26
Norm 7	−0.04	1.7	55	31	29	22
Norm 8	−0.16	1.75	40	32	32	30
Norm 9	0.02	1.6	55	31	28	24
Norm 10	−0.08	1.65	30	34	28	27
Norm 11	−0.2	1.65	75	34	33	28
Norm 12	−0.1	1.65	75	32	30	29
CFL	0.77 (0.48)	1.09 (0.31)	n/a	20 (9)	22 (8)	21 (9)
Norm	−0.09 (0.08)	1.68 (0.06)	55 (17)	27 (8)	27 (6)	28 (2)

 VA measured in LogMAR, CS in log, stereopsis in seconds of arc, and VF in dB. n/a, not available.

Protocol

Each participant completed the protocol wearing their best corrected distance prescription using full aperture trial case lenses in a thin surround trial frame. All participants self-reported no difficulty with walking/negotiating the obstacle while wearing the trial frames. Statistical analyses were performed on the difference in correction between patients' habitual and subjective refraction for the left and right eyes for each group, as differences in magnification/minification can have a significant effect on adaptive gait. ^31,32 Results showed no significant differences between patients' habitual and subjective refraction in either the CFL or normal vision groups for left eye (t[9] = −0.115, P = 0.91 and t[11] = −1.915, P = 0.08, respectively) and right eye (t[9] = −0.426, P = 0.68 and t[11] = 1.149, P = 0.28, respectively). 

Obstacle Crossing

From approximately 5 step lengths away, participants walked up to and stepped over a single obstacle, and continued to walk along the laboratory floor for at least 4 or 5 steps. To minimize the learning effects associated with repeatedly negotiating the same obstacle height, ³³ two obstacle heights (5 and 10 cm) were used. Similar obstacle heights have been used in previous research to reflect typical heights encountered in everyday life. ^34,35 The obstacles (low and high) were constructed from medium density fiberboard (MDF) 1.8 cm thick and 50 cm long. They were light brown in color and presented a high contrast target on the background of navy blue carpet on the laboratory floor to increase the ease of CFL patients perceiving the obstacle and subsequently reducing the risk of tripping. ³⁶  

Before the start of the experiment, participants were asked to complete several familiarization walking trials where they walked along the laboratory with no obstacle present. During this period participants were encouraged to adopt their habitual gait (walking speed/step length) and no data were collected during these trials. During these familiarization trials, masking tape was attached to the floor in the position where the obstacle would be placed during the experiment. Start position was adjusted until the participant's pre-selected lead limb (chosen during the familiarization period and remained the same throughout the experiment) consistently crossed the masking tape during early/mid-swing (5 step lengths from the start position). Obtaining the start position in this manner also ensured that each participant was comfortable with the protocol but did not become familiar with the individual obstacle heights. The masking tape was removed from the floor before the start of the experiment. 

Walking Only

Walking only trials were “nested” within the obstacle crossing trials (presented every third trial; a 1:3 ratio) to avoid participants adopting a repeated motor strategy through repeatedly walking up to and negotiating an obstacle placed at the same distance from their start position. To disrupt somatosensory information further, in the walking only trials participants walked across the laboratory floor starting with their contralateral limb to the obstacle crossing trials. Each obstacle height was negotiated 6 times (obstacle height randomized) so that participants completed 18 trials (12 obstacle crossing and 6 walking only trials). 

3-D kinematic data were collected (at 100 Hz) using a 6 camera 3-D motion capture system (Vicon 460; Oxford Metrics Ltd). Additional information pertaining to the capture system has been described previously. ^37,38 Retro-reflective spherical markers were attached at the following key anatomical locations (placed either directly to the skin or clothing): Superior aspects of the second and fifth metatarsal heads, end of second toes, lateral malleoli, posterior aspect of the calcanei, sternum, and anterolateral and posterolateral aspects of the head. Two additional markers were attached to the upper front edge of the obstacle to determine the height and location of the obstacle within the laboratory coordinate system. Marker trajectory data were filtered using the cross-valedictory quintic spline smoothing routine, with “smoothing” options set at a predicted MSE value of 10 and processed using the PlugIn-Gait software (Oxford Metrics Ltd). 

Data Analysis

Lead and trail-limb foot-off were defined as the instant the resultant vertical and anterior/posterior velocity of each foot's toe marker first increased greater than 200 mm/s for 5 consecutive frames following the period of zero velocity when the foot was planted on the floor. Contact on the floor was defined from when the foot first reduced less than 200 mm/s for 5 consecutive frames (thresholds adopted from published literature³⁵). The following variables have been determined previously as important to assess the kinematics of adaptive gait. ^35,39–41  

Variables Analyzed for Obstacle Crossing

1.  
   Penultimate foot placement – horizontal distance between the lead foot (toe) and obstacle (Fig. 2).
2.  
   Final foot placement – horizontal distance between the trail foot (toe) and obstacle (Fig. 2).
3.  
   Lead and trail limb vertical toe clearance – vertical distance between the upper edge of the obstacle and toe as it crosses the obstacle (Fig. 2).
4.  
   Lead and trail limb vertical heel clearance – vertical distance between the upper edge of the obstacle and the posterior aspect of the calcanei as it crosses the obstacle. This variable was included in addition to vertical toe clearance to provide a complete understanding of how the foot crosses the obstacle. During obstacle crossing, it is possible through dorsiflexing the ankle to increase vertical toe clearance without elevating the foot higher. While this approach subsequently would reduce the risk of the toe contacting the obstacle, it would not reduce the risk of the foot contacting the obstacle as vertical heel clearance in this instance would be reduced.
5.  
   Peak lead and trail limb vertical toe clearance – peak (maximum) vertical distance between the upper edge of the obstacle and the toe during crossing.
6.  
   Lead and trail limb horizontal velocity – horizontal velocity of the toe at the point of obstacle crossing.
7.  
   Foot placement – horizontal distance of the lead and trail foot (toe) at contact with the floor after crossing the obstacle (Fig. 2).
8.  
   Step width – horizontal distance between the lead and trail heel at the instance of lead foot contact on the floor after crossing the obstacle.

Figure 2. 

 

Diagrammatic illustration of foot placement and clearance parameters for the lead and trail feet during obstacle negotiation.

Temporal Variables

1.  
   Double support (DS) time before the obstacle – time from final foot contact on the ground to lead limb toe-off as it is lifted to negotiate/cross the obstacle.
2.  
   Lead limb single support time (SS_Lead) – time spent during obstacle crossing whereby only the trail limb is in contact with the ground.
3.  
   DS time during obstacle crossing (DS crossing) – time from lead limb contact on the floor after crossing the obstacle to trail limb toe-off before crossing.
4.  
   Trail limb single support time (SS_Trail) – time spent during obstacle crossing whereby only the lead limb is in contact with the ground.

Head Flexion

Head flexion included the head angle at penultimate foot contact, final foot contact, peak minimum angle during lead and trail limb single support, and at lead and trail foot contact after crossing the obstacle. Head flexion was normalized to looking straight ahead, such that 0° was looking straight ahead, while positive flexion indicated looking down towards the ground, and negative flexion indicated looking up towards the ceiling. 

Variables Analyzed for Walking Only Trials

For the walking only trials, to ensure participants had achieved their natural walking pattern, and prevent the analysis being influenced by participants accelerating at the start of the trial or decelerating as they reached the end of the laboratory, only one complete step cycle for each limb was calculated as the participant walked across the center of the laboratory. The center of the laboratory coincided with the “0” anterior/posterior coordinate in the Vicon global coordinate system, and penultimate and final foot contact were defined in relation to crossing this threshold. During the walking only trials, the following variables were analyzed: 

1.  
   Average walking velocity – calculated from the sternum between penultimate foot contact up to lead foot contact.
2.  
   Minimum foot clearance (MFC) – defined as the vertical distance between the foot (toe) and floor at peak maximum horizontal velocity of the foot during swing. ^42,43 Increasing MFC during over ground walking has been shown as a cautious gait adaptation to clear the ground safely under conditions of simulated peripheral visual impairment. ⁴⁰
3.  
   Peak velocity during swing – peak horizontal velocity of the foot (toe) during swing.
4.  
   Peak toe clearance – peak (maximum) vertical toe clearance during SS.
5.  
   Step width – medial-lateral distance between the lead and trail heel during DS.
6.  
   Step length – horizontal distance between the lead and trail toe during DS.

Temporal Variables

1.  
   DS time – time spent with both feet in contact with the floor.
2.  
   SS time – time spent with only one limb in contact with the ground.

Head Flexion

Head flexion was defined as head angle at penultimate and final foot contact before crossing, peak minimum angle during SS, and at lead and trail foot contact after crossing. 

Statistical Analysis

Conducting a (separate) initial ANOVA on all the dependant variables for obstacle crossing and walking only trials highlighted that there was no significant effect of trial repetition (P > 0.05). Thus, all data were averaged across repetition. Levene's test for equal variance, and the Kolmogorov-Smirnov test were used to confirm equal variance and normality of the obstacle crossing data (P > 0.05). Obstacle crossing data were analyzed using separate ×2 group (visually impaired, normal) ×2 obstacle height (small and large) mixed between groups repeated measures ANOVA. Level of significance was accepted at P < 0.05 and post-hoc analyses, where appropriate, were performed using Duncan's multiple range. 

Analysis of the walking only data indicated that there were no statistically significant differences in any of the variables calculated between lead or trail limb. Therefore, the data were pooled across limbs for further analysis. Walking only data were analyzed using independent samples t-tests. Head angle at penultimate foot contact, final foot contact, peak minimum head angle during SS, step length, step width, and peak velocity were not distributed normally, and subsequently were analyzed using the Mann-Whitney test. 

Results

Obstacle Crossing

From the 144 obstacle crossing trials completed within the study, the obstacle was only contacted/knocked over on 2 occasions, both by a normal vision participant when negotiating the high obstacle. These trials were discarded from the analysis. 

Penultimate foot placement was unaffected by group (F _1,20 = 2.61, P = 0.12), height (F _1,20 = 1.62, P = 0.22), or interaction effect (F _1,20 = 1.29, P = 0.27). Final foot placement was unaffected by group (F _1,20 = 1.65, P = 0.22), height (F _1,20 = 1.99, P = 0.18), or interaction effect (F _1,20 = 0.26, P = 0.62). DS time before the obstacle also was unaffected by group (F _1,20 = 1.57, P = 0.23), height (F _1,20 = 0.08, P = 0.78), or interaction effect (F _1,20 = 0.31, P = 0.58, Table 3). 

Table 3. 

 

Movement Kinematics during Obstacle Crossing; Group Mean (±SD) across Vision Group and Obstacle Height

	CFL		Normal		ANOVA
	Low	High	Low	High
Penult foot placement (mm)	−824 (139)	−839 (139)	−941 (164)	−941 (166)	n/a
Final foot placement (mm)	−258 (72)	−251 (82)	−294 (39)	−281 (47)	n/a
Lead toe clearance (mm)	218 (58)	216 (43)	170 (34)	189 (26)	G
Lead heel clearance (mm)	207 (51)	212 (41)	137 (44)	154 (52)	G*
Max lead toe clearance (mm)	232 (52)	230 (44)	186 (38)	199 (32)	G
Lead limb horizontal vel (m/s)	3.09 (0.42)	2.99 (0.39)	3.58 (0.53)	3.34 (0.47)	G, H† (g–h)
Step width (mm)	101 (48)	111 (46)	85 (36)	83 (37)	n/a
Lead foot placement (mm)	456 (93)	473 (96)	501 (76)	510 (81)	H
Trail toe clearance (mm)	228 (61)	231 (70)	173 (37)	177 (43)	G
Trail heel clearance (mm)	427 (66)	448 (61)	364 (65)	375 (71)	G
Max trail toe clearance (mm)	237 (58)	244 (58)	183 (40)	194 (44)	G
Trail limb horizontal vel (m/s)	2.64 (0.48)	2.54 (0.37)	3.00 (0.40)	2.74 (0.41)	H
Trail foot placement (mm)	1091 (193)	1116 (188)	1178 (155)	1188 (188)	n/a
Temporal
DS before obstacle (sec)	0.094 (0.044)	0.093 (0.043)	0.068 (0.044)	0.072 (0.045)	n/a
SS lead (sec)	0.639 (0.063)	0.705 (0.072)	0.589 (0.062)	0.650 (0.062)	H†
DS crossing (sec)	0.129 (0.051)	0.1264 (0.047)	0.100 (0.046)	0.109 (0.049)	n/a
SS trail (sec)	0.603 (0.070)	0.622 (0.043)	0.582 (0.057)	0.618 (0.067)	H†
Head Flexion
Penult foot contact (°)	10 (7)	10 (7)	2 (9)	3 (13)	n/a
Final foot contact (°)	15 (7)	15 (9)	3 (10)	4 (13)	G
Min head angle SS lead (°)	18 (12)	17 (10)	5 (11)	6 (11)	G
Lead foot contact (°)	11 (13)	12 (11)	−1 (10)	−1 (10)	G
Min head angle SS trail (°)	9 (12)	11 (12)	−2 (11)	−3 (10)	G
Trail foot contact (°)	3 (9)	5 (9)	−6 (10)	−7 (10)	G

 Data illustrate the effects of vision group (G) and obstacle height (H). Factors found to be significant are shown by capital letter (P < 0.05), asterisks (* P < 0.01), and dagger accent († P < 0.001), and interactions between factors are denoted by lower case letter in brackets. N.B, negative foot placement values indicate placement before the obstacle; n/a, no significant effect.

Lead limb vertical toe and heel clearance was significantly higher in CFL patients compared to normal vision patients (F _1,20 = 4.87, P = 0.04 and F _1,20 = 10.19, P = 0.005, respectively), but was unaffected by height (F _1,20 = 2.34, P = 0.14 and F _1,20 = 3.00, P = 0.10, respectively). Although visual normals increased lead limb vertical toe clearance when negotiating the high compared to low obstacle, there was no significant interaction effect (F _1,20 = 2.96, P = 0.10, Fig. 3a). Trail limb vertical toe and heel clearance was significantly higher in CFL patients compared to normal vision patients (F _1,20 = 6.17, P = 0.02 and F _1,20 = 6.09, P = 0.02, respectively, Fig. 3b); however, it was unaffected by height (F _1,20 = 2.66, P = 0.61 and F _1,20 = 2.62, P = 0.12, respectively) and there was no significant interaction effect (F _1,20 = 0.00, P = 0.96 and F _1,20 = 0.25, P = 0.62, respectively). 

Figure 3. 

 

Group mean (± SE) lead (a) and trail (b) limb vertical toe clearance between CFL and normal vision patients when negotiating the low and high obstacles. There was a significant main effect for group in lead and trail toe clearance (P < 0.04).

Lead limb horizontal velocity was affected significantly by group (F _1,20 = 4.43, P = 0.049) and obstacle height (F _1,20 = 32.77, P = 0.000), and there also was a significant group-by-obstacle height interaction (F _1,20 = 5.47, P = 0.03). Post-hoc testing showed that CFL patients had a significantly lower velocity when negotiating the low obstacle compared to visual normals; however, it was not significant for the high obstacle. Visual normals had significantly lower velocity when negotiating the high compared to low obstacle (Fig. 4). Trail limb horizontal velocity was reduced significantly when negotiating the high compared to low obstacle (F _1,20 = 6.91, P = 0.017, Table 3). There was no significant group or interaction effect (F _1,20 = 2.64, P = 0.12 and F _1,20 = 1.38, P = 0.26, respectively). 

Figure 4. 

 

Group mean (±SE) lead limb horizontal velocity for CFL and normal vision patients when negotiating the low and high obstacles. A significant group-by-obstacle height interaction was observed (P = 0.033).

Peak lead limb vertical toe clearance was significantly higher in CFL patients compared to normal vision patients (F _1,20 = 5.17, P = 0.035), but was unaffected by obstacle height (F _1,20 = 0.92, P = 0.35, Table 3). There was no significant interaction effect (F _1,20 = 1.80, P = 0.20). Peak trail limb vertical toe clearance was significantly higher in CFL patients compared to normal vision patients (F _1,20 = 6.42, P = 0.02); however, it was unaffected by height or significant interaction effect (F _1,20 = 1.70, P = 0.21 and F _1,20 = 0.07, P = 0.80, respectively). 

SS_Lead time was significantly longer when negotiating the high compared to low obstacle (F _1,20 = 49.27, P = 0.001) and showed a trend (F _1,20 = 3.87, P = 0.06) of being longer in CFL patients compared to visual normals (Table 3). There was no significant interaction effect (F _1,20 = 0.06, P = 0.81). SS_Trail time was significantly longer when negotiating the high compared to low obstacle (F _1,20 = 14.67, P = 0.001). There was no significant group or interaction effect (F _1,20 = 0.23, P = 0.64 and F _1,20 = 1.39, P = 0.25, respectively). 

Lead foot placement was significantly affected by obstacle height (F _1,20 = 4.84, P = 0.04) and was positioned further in front of the obstacle when negotiating the high compared to low obstacle. There was no significant group or interaction effect (F _1,20 = 1.00, P = 0.33 and F _1,20 = 0.40, P = 0.54, respectively). Trail foot placement was not affected by group (F _1,20 = 0.95, P = 0.34) or obstacle height (F _1,20 = 2.76, P = 0.11); there was no significant interaction effect (F _1,20 = 0.54, P = 0.47). DS crossing time was not affected by group (F _1,20 = 1.29, P = 0.27), obstacle height (F _1,20 = 0.19, P = 0.67), or interaction effect (F _1,20 = 0.79, P = 0.39). Step width was not affected by group (F _1,20 = 1.61, P = 0.22) or obstacle height (F _1,20 = 0.45, P = 0.51), and there was no interaction effect (F _1,20 = 0.87, P = 0.36, Table 3). 

There was a trend (F _1,20 = 3.90, P = 0.06) of head angle at penultimate foot contact being flexed increasingly positively in CFL patients compared to visual normals. There was no effect of obstacle height or significant interaction effect (F _1,20 = 0.02, P = 0.89 and F _1,20 = 0.15, P = 0.71, respectively). Head angle was significantly greater (positively flexed) in CFL patients compared to normals at final foot contact (F _1,20 = 6.72, P = 0.02), during lead limb SS (F _1,20 = 5.926, P = 0.03), lead foot contact (F _1,20 = 7.10, P = 0.02), trail limb SS (F _1,20 = 6.85, P = 0.02), and at trail foot contact (F _1,20 = 6.50, P = 0.02, Fig. 5a). There was no effect of obstacle height or significant interaction effect in any of the aforementioned head angle variables (P > 0.05, Table 3). 

Figure 5. 

 

Group mean (±SE) head flexion during penultimate (Penult) and final (Final) foot placement before crossing, peak minimum head flexion during lead and trail limb SS (SS lead, SS trail), and lead and trail foot contact (Lead cont, Trail cont) after crossing the obstacle for CFL and normal vision patients. (a) Obstacle crossing trials. (b) Walking only trials. 0° flexion indicates looking straight ahead and large positive head flexion indicates looking down towards the floor. N.B, SS lead, and trail data collapsed together for walking only trials.

Walking Only Trials

From the 13 dependant variables analyzed, there were no significant differences between the two vision groups. The critical statistics and means for each variable can be found in Table 4. Head flexion values at key points during the walking only trials can be found in Figure 5b. 

Table 4. 

 

Movement Kinematics during Walking Only Trials; Group Mean (±SD) across Visual Group

	CFL	Norm	Critical Statistic
Walking velocity (m/s)	1.159 (0.20)	1.21 (0.21)	t(20) = −0.54, P = 0.59
MFC (mm)	59 (13)	61 (15)	t(20) = −0.89, P = 0.68
Peak velocity swing (m/s)	4.28 (0.51)	4.49 (0.43)	U = 44.00, z = −0.71, P = 0.48
Peak toe clearance (mm)	114 (24)	125 (23)	t(20) = −1.13, P = 0.30
Step width (mm)	98 (53)	80 (35)	U = 42.00, z = −0.85, P = 0.42
Step length (mm)	668 (79)	719 (75)	U = 30.00, z = −1.71, P = 0.10
Temporal
DS time (sec)	0.065 (0.041)	0.053 (0.039)	t(20) = −0.68, P = 0.51
SS time (sec)	0.507 (0.060)	0.540 (0.059)	t(20) = −1.25, P = 0.75
Head Flexion
Penult foot contact	−6 (9)	−11 (8)	U = 43.00, z = −0.78, P = 0.43
Final foot contact	−5 (9)	−11 (7)	U = 31.00, z = −1.64, P = 0.10
Peak min SS	−3 (9)	−9 (7)	U = 33.00, z = −1.49, P = 0.10
Lead foot contact	−4 (9)	−11 (7)	t(20) = −1.95, P = 0.71
Trail foot contact	−5 (8)	−11 (7)	t(20) = −1.90, P = 0.61

Discussion

The aim of our study was to investigate (through detailed kinematic analysis) how patients with CFL complete an adaptive gait task of obstacle negotiation and level walking when compared to normal vision participants of a similar age, and determine whether task difficulty has a significant effect on performance. Participants were tasked with walking up to and stepping over an obstacle during ongoing gait and walking across a room with no obstacle present. Our results demonstrated that when completing the obstacle crossing task, compared to visual normals, patients with CFL adopted a cautious crossing strategy to reduce the risk of contacting the obstacle and tripping/falling. There was no significant difference in the kinematics of gait when completing the walking only trials between groups. These findings indicate that patients with CFL adopt a cautious gait strategy during tasks that present a high risk of falling, such as obstacle crossing. 

Obstacle Crossing

Compared to visual normals, when patients with CFL negotiated the obstacle, they lifted their lead and trail foot significantly higher to avoid the obstacle. This was evidenced through a greater vertical lead/trail toe and heel clearance, and peak maximum vertical lead/trail toe clearance (Table 3 and Fig. 3). Such adaptations in the movement indicate a cautious stepping strategy to reduce the risk of contacting the obstacle, which is most likely due to patients being unable to acquire a precise visual representation or accurately perceive the obstacle's characteristics due to their degraded vision/CFL (Table 2). This cautious stepping strategy also has been reported in response to negotiating an obstacle or step under conditions of correctable (cataract) and simulated visual impairment. ^{25,34,44–47} A cautious gait strategy also was reported in patients with CFL when walking across different surface terrain. ¹⁵ At the point during the movement when the lead foot (toes) were over the front edge of the surface, patients with CFL had adopted a more dorsiflexed ankle (toes were higher than the heel), whereas for visual normals the toes were lower than the heel. ¹⁵ An increasingly dorsiflexed ankle will reduce the risk of catching the toes on the surface edge and subsequently tripping. However, without elevating the entire foot higher, the heel still will be at risk of contacting the surface/obstacle edge. The stepping strategy adopted by patients with CFL in our study was an overall “safer” response to minimize the risk of foot contact compared to that shown by Spaulding et al. ¹⁵ These differences in stepping strategy between the current and previous research ¹⁵ may be attributed to the different task employed. 

Increased vertical toe clearance is not the only variable that has been shown to indicate a safer stepping/crossing strategy. Indeed, contacting a surface edge with the swinging limb that has a higher horizontal velocity increases the risk of tripping. ⁴⁸ Compared to visual normals, patients with CFL had a significantly lower lead limb horizontal velocity when negotiating the low obstacle, and lower (but not significantly different) horizontal velocity when negotiating the high obstacle (Fig. 4). These results, in combination with vertical foot clearance parameters highlighted further the cautious crossing strategy adopted by patients with CFL when negotiating the obstacle. It also is relevant to note that visual normals had significantly reduced lead limb horizontal velocity when negotiating the high compared to low obstacle (Fig. 4). This is consistent with previous research highlighting that normal vision participants actively reduced horizontal toe velocity when negotiating increasingly higher obstacles to reduce the risk of tripping if contact occurred. ⁴⁸ However, there was no significant height effect in patients with CFL, which may indicate that irrespective of obstacle height, patients purposefully adopted a low crossing velocity/cautious crossing behavior to minimize the risk of tripping. This, however, requires confirmation with further research. The reduced lead limb horizontal velocity among patients with CFL and the increased vertical lead foot clearance most likely contributed to the nonsignificant increase in SS_Lead time (P = 0.06) in patients with CFL compared to visual normals (Table 3). Increasing SS_Lead time provides greater opportunity during the movement to initiate “online” corrections to “fine tune” lower limb trajectory and reduce the risk of contacting the obstacle. ^49,50 On the other hand, increasing the time spent during single support also may have a negative influence on dynamic stability, increasing the risk of falling, as shown in patients with simulated visual impairment when negotiating a single step. ⁵¹  

Some previous research also has highlighted that during adaptive gait, under conditions of simulated visual impairment, patients adopt a more stable gait strategy during single support through reducing medial-lateral movement of the center of pressure, ensuring it remains closer to the base of support. ²⁵ In our study, this variable (and others associated with the center of pressure) were not analyzed due to the difficulties associated with data collection by ensuring only one (entire) foot lands on the force plate immediately before stepping over the obstacle during ongoing gait. The analysis of the center of pressure can be measured accurately in adaptive gait tasks performed from a stationary standing position (which was the case for Heasley et al.²⁵) but not so easily during ongoing gait as in our study. 

Patients with CFL had significantly greater head flexion during the approach to and step over the obstacle (Fig. 5a), except at penultimate foot contact, in which a trend was observed (P = 0.06). When completing the walking only trials, there was no significant difference in head flexion between patients with CFL and normal vision (Fig. 5b). These results suggested that patients with CFL only increase head flexion to look down at more immediate areas of the floor, to increase visual input, when the ground characteristics become more demanding. ^15,16 This strategy was reported similarly by Spaulding et al. when patients with CFL were compared to visual normals when negotiating complex terrain or experienced extreme lighting conditions during level walking. ^15,16 While increasing head flexion may have resulted in increased visual sampling of the immediate floor area, without recording participant's gaze behavior we cannot be certain; clearly this needs further investigation. 

During obstacle crossing, the foot must be lifted sufficiently high over the obstacle to ensure minimal risk of contact. Increasing the height the foot is lifted above the obstacle requires greater “work” from the individual, thereby resulting in a greater potential energy cost. This creates a conflict between lifting the foot sufficiently high to ensure minimal chance of obstacle contact and also ensuring the conservation of energy. ²⁵ Previous research has highlighted that this conflict is resolved through decreasing the height the foot clears the obstacle/step as individuals become more familiar with the task. ^25,34,52 In our study, task familiarity (or trial repetition) had no effect on foot clearance parameters in CFL patients (or any other dependant variable analyzed). Unlike previous research, our study recruited patients with actual rather than simulated visual impairment. This may indicate that while patients with simulated visual impairment (through cataract or visual field loss^25,34,52) use sensory information/feedback to update subsequent repetitive movements and conserve energy, patients with actual visual impairment consistently prioritize the requirement for safety over the need for conservation of energy during adaptive gait. It also is relevant to note that Heasley et al. ²⁵ and Simoneau et al., ⁵² who tasked patients with step negotiation, reported that participants would have received additional feedback from somatosensory information pertaining to the height of the step (after landing, which would update subsequent responses). This information obviously would not be available when stepping over an obstacle. 

Compared to visual normals, patients with CFL had no significant difference in final foot placement before the obstacle, but significantly increased vertical foot clearance over the obstacle (Table 3). This indicates that patients with CFL had little difficulty locating the for/aft position of the obstacle, but were unable to determine its height accurately. Patients with CFL had either no or minimal stereopsis (Table 2), which limits depth perception. Previous research has shown that stereopsis is an important visual cue for making accurate judgments about surface height but not location. ^44,53,54 While our results support this previous research, it also is possible that among our patients with CFL these findings are attributed to the overall reduced VA, CS, and/or VF loss. 

Irrespective of vision group, when negotiating the high obstacle, the trail foot was positioned further in front of the obstacle after crossing. The precise reason for this finding remains unclear. This may be attributed to the nonsignificant decrease in final foot placement (the distance the foot was positioned from the obstacle) before negotiating the high compared to low obstacle. 

Walking Only

When patients with CFL were tasked with completing the walking only trials, compared to patients with normal vision there was no significant difference in any of the dependant variables analyzed (Table 4). These results support previous research that found no difference in the time it took patients with CFL to complete a mobility course or walk a pre-specified distance compared to participants with normal vision. ^13,18,19 Results from our study and aforementioned mobility research contrast with previous research, which highlighted that when completing a mobility course, CFL was associated most strongly with reduced walking speed and increased number of bumps. ¹² We hypothesize that this likely is attributed to the complexities of the obstacle course used. However, differences between the current and previous research also may be attributed to adaptation to vision loss. ³⁷  

Task Difficulty

In our study, results from the obstacle crossing and walking only trials suggested that task difficulty has a significant effect on how patients with CFL complete everyday mobility tasks when compared to visual normals. Indeed, in complex adaptive gait situations, such as obstacle crossing, patients with CFL adopted a cautious crossing strategy to reduce the risk of tripping on the obstacle, whereas in the less demanding situation of level walking there was no difference in the kinematics of gait between groups. In support of this, Kuyk and Elliott also found that the mobility performance of patients with CFL was significantly worse when walking in increasingly difficult environments. ²⁰  

It is important to highlight that while our results can be generalized across patients with CFL who are fit and healthy, among patients with CFL who have a number of additional comorbidities (such as using a mobility aid, are a repeated faller, and so forth), it is possible that they will adopt a cautious walking strategy across all gaits tasks when compared to visual normals, and may even have a different gait strategy compared to healthy patients with CFL. This has been suggested by previous research highlighting that normal vision older adults with a “high risk” of falling adopt a stepping strategy that contributes to them being at a higher risk of tripping and falling compared to “low risk” fallers of a similar age. ⁵⁵  

Conclusions

When patients with CFL were tasked with stepping over an obstacle during ongoing gait, compared to visual normals, they lifted their lead and trail foot higher to cross the obstacle, reduced lead horizontal foot velocity (only significant when crossing the low obstacle) and increased head flexion to look down at more immediate areas of the floor/obstacle. These changes indicated that during obstacle negotiation, patients with CFL adopt a cautious stepping strategy to reduce the risk of contact and tripping/falling. When completing the walking only trials, there was no significant difference in the kinematics of gait between patients with CFL and visual normals. Collectively, these results suggested that compared to visual normals, patients with CFL adopt a cautious gait strategy only during tasks that present a high risk of falling. 

Acknowledgments

Daryl Tabrett conducted the visual assessments on the participants and assisted during data collection. Amy Scarfe assisted during data collection and processing. 

Footnotes

 Disclosure: M.A. Timmis, None; S. Pardhan, None